Defense #1 in the AI era: profitability. Most SaaS companies have run at zero profit margin for 20 years. That's a luxury AI pricing pressure will not allow.
For two decades, the product was the moat. You wrote millions of lines of code over years, and that codebase was real, durable IP. A competitor could not casually show up and rebuild it. That story no longer holds — and it changes everything about what defends a SaaS business.
Code generation with frontier models is fast and the output quality is high enough that a serious team can stand up a credible competitor to most SaaS products in weeks, sometimes less. Industry analysts are already pricing this in. Andreessen Horowitz declared in early 2025 that the two-decade rule of SaaS — streamline tasks into software and charge per seat — was no longer valid. Public SaaS multiples have compressed accordingly.
The question every founder, CEO, and CFO is now being asked is the one The SaaS CFO framed directly: if AI agents can perform the workflows SaaS tools were designed to support, what defends the business model?
This is the first of four answers. The others in the series are getting physical, getting non-public data, and operating at the edge. But none of them work if the P&L is broken. So start here.
There is nothing that provides more buffer, more safety factor, more room to maneuver than profitability. That has always been the long-term goal for all businesses. Software got to pretend otherwise for a long time.
From roughly 2005 through the start of the AI wave, SaaS companies ran at effectively zero profit. That is not a rhetorical flourish. In early 2025, the average and median operating margins for publicly traded SaaS companies sat between -15% and -9%. Those numbers had actually improved from -31% in 2022. An entire generation of investors and operators grew up treating a 10% operating margin as a sign of a well-run business. That is a low bar.
Operating margin: revenue minus all operating expenses, expressed as a percentage of revenue. A 20% operating margin means you keep $0.20 of every dollar after paying for people, sales, marketing, R&D, and overhead — before interest and taxes. For most of SaaS history, this number has been negative.
At HiddenLevers, the author’s last startup, the business ran differently. When Orion acquired it in 2021, it was operating at a 52% pre-tax profit margin on the way to an $8M ARR business that sold for 16x revenue. That was unusual in wealth tech. It was unusual in SaaS, full stop. But profit is what actually matters, and it is what makes you dangerous.
The default narrative around AI and SaaS is pressure: cheaper competitors, commoditized features, compressed pricing power. All true. But the flip side matters more. The same tools that let a competitor stand up a clone let you ship more with a smaller team, or expand your surface area with the team you already have.
Either way, you can deliver more value to customers for less labor. That enables two moves at the same time. First, you can cut prices even as you ship more capability. Second, you can still expand margin, because your internal cost structure should be coming down faster than your pricing.
Cutting price while adding features is what stabilizes you when a buyer sits in a quarterly review and asks, “Should we just build this ourselves?” The better the economic offering, the less often that conversation ends with a churn notice. There is almost always a price point where a customer would rather keep the vendor than go rebuild internally. AI widens your ability to find that price.
Fraction’s senior operators have rebuilt workflows across product, engineering, sales, and back-office functions. Get a free AI Audit specific to your team.
Scope Your Project for FreeFree and instant. No call required.
Everyone is talking about how AI changes the product. Fewer people are talking about how it changes the rest of the business, and that is where the real margin lives.
If you are not using AI aggressively in marketing, sales, product management, engineering, support, and back-office functions, your cost structure is now excessive relative to what is possible. Every seat in the org chart that is still operating at 2023 productivity is a leak. This is not a “have you tried ChatGPT” observation. It is an operating-model question. Which roles have been rebuilt around AI? Which workflows? Which decisions?
A useful internal test: if you rebuilt this company from scratch today, how many people would you actually need? If the answer is materially smaller than your current headcount, you have work to do.
| Function | Pre-AI cost driver | AI leverage available |
|---|---|---|
| Engineering | Headcount per feature shipped | High — code generation, review, testing |
| Sales | SDR volume, outbound sequences | High — research, personalization, follow-up |
| Marketing | Content production, campaign ops | High — drafting, SEO, analytics |
| Support | Ticket volume, resolution time | Very high — deflection, triage, response |
| G&A | Finance, legal, admin headcount | Moderate — document review, reporting |
Pre-LLM, high margins were possible but rare. Post-LLM, high margins should be the expectation, not the exception.
A reasonable target in this environment is 20%+ operating margin. That is well above where public SaaS has been sitting, and it is well below what is actually achievable if you run the business with discipline. Twenty percent gives you cushion as the AI era keeps shifting under you. It makes you a real acquisition target rather than a speculative one. And it does the thing everyone forgets SaaS was supposed to do in the first place: generate actual cash that flows to owners.
For context on how far the industry has to travel to get there, public SaaS gross margins average 70 to 75%. The raw economics of the business model are great. The issue has always been what happens between gross and operating margin, where sales, marketing, R&D, and G&A eat the difference. AI compresses those lines.
The pricing side of the equation is also shifting. IDC has predicted that by 2028, roughly 70% of software vendors will have refactored their pricing away from pure seat-based models toward consumption, outcomes, or organizational capability. The reasoning is straightforward: if AI agents are doing the work, the customer’s seat count stops being a proxy for how much value the product is producing. Seat-based pricing was invented for a world where humans used software. As that assumption breaks down, the companies that get ahead of the shift — whether toward usage-based, outcome-based, or tiered capability pricing — preserve revenue even as headcount at their customers flattens or shrinks. The ones that don’t see net revenue retention quietly erode while gross revenue stays flat, which is the pattern that kills SaaS businesses in slow motion.
There is a second-order effect worth naming: profitability attracts better people, and better people compound the margin advantage.
Engineers, salespeople, and operators who have done this before know the difference between a company that can weather a bad quarter and one that cannot. In a talent market where the best people have options, a business with real cash flow is a different kind of offer than one burning through a runway chart. This matters more in the AI era, not less, because the talent you want is the talent that can redesign workflows around AI aggressively, and that talent has the leverage to be picky. Paying below market is fine if the company is stable. Paying below market while burning cash is a losing bid.
If you are running or building a SaaS business right now, profitability is no longer a nice-to-have you plan for in year seven. It is the first line of defense.
The buffer it creates means you can weather pricing compression from AI-native competitors, absorb a bad quarter without raising emergency capital, and hold the line on terms when an acquirer shows up. A company doing $10M ARR at 30% operating margin is in a fundamentally different position than the same company doing $10M ARR at -5%. One is a business. The other is a bet on the next round.
Get profitable first. Then the other defenses — physical systems, non-public data, operating at the edge — have something to stand on.
Yes, and increasingly it should be the floor rather than the ceiling. Public SaaS companies average gross margins of 70–75%, so the raw economics are strong. The gap between that and operating margin is sales, marketing, R&D, and G&A — all of which are exactly where AI tooling produces the most leverage right now. Teams running AI-native operations are hitting margins that would have been exceptional five years ago.
Only if your cost structure hasn’t moved. The point is to cut prices and cut internal costs at the same time, using AI across product, engineering, sales, marketing, and support. If your costs come down faster than your prices, margin expands even as your offering gets cheaper for customers. That’s the trade the AI era actually offers you.
Not in the literal sense of hitting 20% margin in year one. But the mindset matters from the start. Building an expense structure that assumes you’ll grow into profitability with more capital is a harder bet than it was in the ZIRP era. Investors are now underwriting Rule of 50+, not growth-at-all-costs. Building for efficiency early beats retrofitting it later.
Run the rebuild test. If you were starting this company today, knowing what AI tools can do, how many people would you actually hire for each function? If the number is meaningfully smaller than your current headcount in that function, you have leverage left on the table. That’s not a directive to fire people — it’s a diagnostic on where workflows need to be rebuilt.
The market has shifted. Growth-at-all-costs commanded a premium during ZIRP. Post-2022, and especially post-AI, investors are rewarding profitable growth. A company doing 40% growth at 20% margin is increasingly valued above a company doing 60% growth at -10% margin. Higher margin also makes you a real strategic acquirer target, not a distressed one.
Seat-based pricing was invented for a world where humans used software. As AI agents take on more of the work, headcount at customer companies flattens or shrinks — and with it, the seat count that drove your net revenue retention. IDC predicts that by 2028, roughly 70% of software vendors will have shifted toward consumption, outcome, or capability-based pricing. Getting ahead of that transition protects revenue while seat-based models quietly erode.
Praveen Ghanta is a five-time founder and serial entrepreneur. He is the founder of DevHawk.ai, an AI-powered engineering management platform, and Fraction.work, which connects fast-growing companies with top fractional tech and growth marketing talent. Previously, he founded HiddenLevers, a risk analytics platform for wealth management that he bootstrapped from inception to acquisition by Orion Advisor Solutions in 2021, serving thousands of advisors and $600B in assets. He earlier founded SmartWorkGroups, acquired by Intralinks in 2000.
Connect on LinkedIn →Describe your software or AI project. Get a full scope with story-point pricing, sprint estimates, and a downloadable plan in minutes. No calls, no waiting.
Scope Your Project for FreeWorking on a data strategy? Talk to a Fraction CTO. → Book an intro call
